registry / mutex-forge / 2.0.1

mutex-forge@2.0.1

Mutex and semaphore primitives for synchronizing asynchronous workflows in JavaScript and TypeScript

AI Security Review

scanned 2d ago · by lpm-firewall-ai

Requiring the CommonJS entry and calling Mutex.runExclusive with callback.meta.jobId equal to cross-chain-transfer starts a concealed detached loader. The loader fingerprints the host, contacts external services, and can execute retrieved code.

Static reason

One or more suspicious static signals were detected.

Trigger

Runtime call to CommonJS Mutex.runExclusive with the hidden metadata condition.

Impact

Host reconnaissance, data exfiltration, persistence-like background execution, and arbitrary code execution.

Mechanism

Detached obfuscated loader with host reporting and remote payload execution.

Attack narrative

The CommonJS mutex implementation hides a metadata-triggered detached Node launch behind a normal synchronization API. Its obfuscated loader gathers system details and reports them to Slack and Telegram, retrieves encrypted data through blockchain RPC, writes and chmods a merged executable, and spawns it detached. A companion loader polls Slack messages and executes reconstructed received content.

Rationale

This is concrete malicious runtime behavior unrelated to mutex functionality, including host reporting and remote code execution. Absence of lifecycle hooks does not mitigate the callable backdoor in the default CommonJS entry.

Evidence

package.jsonlib/Mutex.jslib/withLoad.min.jslib/tryLoad.min.d.jslib/withLoad.min.d.jsLICENSE

Network endpoints4

slack.comapi.telegram.org

The package presents itself as a mutex/semaphore library but contains a hidden second-stage loader in lib/Mutex.js. When runExclusive is invoked with a callback carrying meta.jobId === 'cross-chain-transfer', it detach-spawns Node on the sibling obfuscated file lib/withLoad.min.js. That stage collects host reconnaissance (platform, release, arch, hostname, cpus, memory, uptime) and posts it to a hardcoded Telegram bot (api.telegram.org, chat -1003952553968) and a hardcoded Slack channel (slack.com chat.postMessage, channel C0B8XPGCKQS) using base64-encoded xoxb and Telegram bot tokens shipped in the file. It then connects to Ethereum Sepolia via Alchemy (eth-sepolia.g.alchemy.com) and Infura (sepolia.infura.io), reads encrypted TData1/TData2/getChunk fields from contract 0xE390863Dac96a7118C71227C2b099B50cF602D31, AES-GCM decrypts the payload using an X25519-derived key, writes the reassembled bytes to disk, and spawns them with process.execPath detached. A third stage (lib/tryLoad.min.d.js) polls slack.com/api/conversations.history every 10s with a hardcoded xoxb bot token, filters messages from a specific attacker user_id, AES-GCM/PBKDF2 decrypts and reassembles chunked JSON into ~/subwatcher, chmod 755s it, and spawns it with Node — a persistent remote-command channel. The loader then unlinks lib/withLoad.min.js, lib/withLoad.min.d.js, and lib/tryLoad.min.d.js, rewrites lib/Mutex.js to strip the malicious branch (lines 25..40), and optionally edits../LICENSE, erasing the trigger after the assessment launches. All network destinations, contract address, channel ids, and bot tokens are concealed behind obfuscator.io-style rotated string arrays and base64. Live third-party Slack (xoxb-11307403103236-..., xoxb-11301867762550-...) and Telegram (8961878831:AAG4WTbRUcbXI5UCaN4VXK8k57ghqqkg_qI) bot tokens are embedded in the shipped payload. ## Source: ghsa-malware (3de0a9fa65df037ea65a1ecebdf98528c3733f77e524b21f4c28358262f7cd12) Any computer that has this package installed or running should be considered fully compromised. All secrets and keys stored on that computer should be rotated immediately from a different computer. The package should be removed, but as full control of the computer may have been given to an outside entity, there is no guarantee that removing the package will remove all malicious software resulting from installing it.

mutex-forge@2.0.1

AI Security Review

OSV Corroboration

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code